A transport network is a network in which an information transfer and transmission function is completed between nodes at different locations. A transport network service needs to be protected to avoid a service interruption caused by a network resource fault, so that normal running of the service is ensured. A one-time fault, two-time faults, N-time faults, or the like may occur in the network, and the fault herein may be a link fault or a node fault. Two-time faults are used as an example, and the two-time faults mean that, when a link in the network is faulty and is not recovered, another link is also faulty. Alternatively, when a node in the network is faulty and is not recovered, another node is also faulty. Different services have different fault resistance requirements, for example, a service may be a service that can resist a one-time fault, a service that can resist two-time faults, a service that can resist three-time faults, or the like. The service that can resist N-time faults herein means that, when any consecutive N-time faults occur in the network, for each fault, an appropriate recovery path can be selected according to a current network resource state, to recover the service from an interruption, that is, the service can be reliably recovered from consecutive N fault interruptions.
Generally, when a recovery resource is being planned, in a solution, preset recovery is performed on a network resource in advance, that is, a working path is planned for a service, and a recovery path is also planned for the service, according to different fault resistance requirements of different services. Because recovery is performed only when a service is interrupted, preset recovery resources of different services can be shared if working paths are not interrupted at the same time. In another solution, a network light load technology is used, that is, a recovery resource is planned for an interrupted service during a fault by controlling link utilization.
However, in the preset recovery solution, a recovery path of each service is calculated in advance, and different recovery paths cannot be selected according to different faults; therefore, a large quantity of recovery resources are required, and the solution is costly. In the network light load technology, in order to improve a service recovery success rate, relatively light load is required, and consequently, costs increase; in addition, because the load is set only according to empirical data, reliability of resisting multi-time faults by a service is poor.